85 replies to this topic

[Lazy]

If the question you want answered is 'how quickly can this car get round this track this weekend' then I don't see why you would want to exclude any lap from the sample.

Maybe if it doesn't shed your driver in the best light?

[RubalSher]

Absolutely, and this, in a nutshell, is (IMHO) the main intended purpose of this metric/statistic.

 

I was thinking about this more and since yours is a cumulative approach, the one thing it would give a good indication of at the end of the season is to rank cars from fastest to slowest over the entire year. And if somebody is following the % numbers, he will get an idea of whether the gap increased or reduced over the year.

 

Is this what the numbers would eventually show?

[karlth]

If the question you want answered is 'how quickly can this car get round this track this weekend' then I don't see why you would want to exclude any lap from the sample.

 

Would you include laps from FP1 in China when Alonso was faster than the Mercedes'?  Compare the laps from the race with the laps done in wet qualifying?  Would you include Gutierrez' best lap from the race or the best lap from the Lotus?

 

Of course not because the conditions and configurations are vastly different between the cars.

 

Qualifying is driven with the cars at maximum performance in the same conditions (not necessarily the best conditions).  Why would you want to muddle the water with laps that are not comparable?

[Dolph]

What on earth is this supposed to show us?  Who was the last driver to put on fresh tires?  What driver didn't need to push?

 

What is qualifying not supposed to tell us?

 

Mercedes is the fastest and Marussia + Caterham trailing last.  I'd say its a good ballpark comparison of who are the fastest drivers in current cars. If that's smth you can't understand then this thread is not for you.
 

[RubalSher]

Would you include laps from FP1 in China when Alonso was faster than the Mercedes'?  Compare the laps from the race with the laps done in wet qualifying?  Would you include Gutierrez' best lap from the race or the best lap from the Lotus?

 

Of course not because the conditions and configurations are vastly different between the cars.

 

Qualifying is driven with the cars at maximum performance in the same conditions (not necessarily the best conditions).  Why would you want to muddle the water with laps that are not comparable?

 

From each race, only one lap is chosen and that is the fastest set by the driver over the entire weekend.

 

Fastest lap is usually set in ideal condition for each car, so what is the problem here.

[karlth]

Fastest lap is usually set in ideal condition for each car ...

 

No that is a fallacy.  The cars are in optimal configuration during qualifying.  During free practice cars are running in different configurations, even between drivers in the same team.

 

If qualifying is washed out then you can end up comparing/analyzing fastest laps of cars running light fuel loads on fresh tires with cars running heavier and on used tires.

[RubalSher]

No that is a fallacy.  The cars are in optimal configuration during qualifying.  During free practice cars are running in different configurations, even between drivers in the same team.

 

If qualifying is washed out then you can end up comparing/analyzing fastest laps of cars running light fuel loads on fresh tires with cars running heavier and on used tires.

 

Yes we know for sure they are running in optimal configuration in qualy.

 

But the fastest times set in free practice (if qualy is wet) are also almost always very close to this configuration. For example, option tyres, low fuel runs, fresh tyres, no traffic. If you have been seeing free practice, all cars do the optimal runs in free practice as well.

[karlth]

Yes we know for sure they are running in optimal configuration in qualy.

 

But the fastest times set in free practice (if qualy is wet) are also almost always very close to this configuration. For example, option tyres, low fuel runs, fresh tyres, no traffic. If you have been seeing free practice, all cars do the optimal runs in free practice as well.

 

Far from it.  For the last 4 years Red Bull for example have rarely run in qualifying configuration until Q2.

[RubalSher]

Far from it.  For the last 4 years Red Bull for example have rarely run in qualifying configuration until Q2.

 

But that applies to everyone. Nobody goes all out in practice sessions either. For the most part, you will be getting the qualy data anyway, so your concern should be alleviated. Just that 3 of the first 4 qualys were wet and data was likely taken from practice sessions.

 

Ride with it for now, I think you got the idea.

[karlth]

[Coops3]

You guys are a tough lot.

It was an exercise that Gary Anderson always provided in the Autosport magazine race previews (albeit with nice pictures of the cars showing the relative ditsance between them on the page) and which I enjoyed following.

To me, this information provides a relative position on the speed differential between each of the cars and it does also give a general position on the relative position between the team drivers. Using 100% as the baseline, equates to a 1min 40sec lap, so over that lap length (slightly longer than the avg), a 1% variation equates to a 1 sec per lap variation.

It also as Imateria says, provides a good basis to show the development improvement, or lack of, for each team over the season. Later on I will look to show it in graph form which will help show whether the teams are getting closer or further away from the ultimate pace.

Haha, sorry dude. I appreciate the effort!

[rooksby]

It's cargo-cult statistics. It has got numbers, and it mechanically processes them, and it looks scientific, and something gets pumped out the end appearing all very precise ... and none of that translates to it being useful, informative, reliable, or very much worth doing.

 

The problem is not in thinking of counter-examples of typical race-weekend situations that easily make a mockery of this methodology, the problem would be trying to stop once you got started.

 

Two pages of commentary here, none of it so far about any insights offered up by this dataset. If you enjoy picking an arbitrary datapoint, comprehensively stripping it off any and all context, smearing it with an average function and then putting it in a table to confuse it with information then this is definitely the place to be. The procedure illuminates little of any great significance, any conclusions such that they are lack confidence, and the process seems to spend much of its time actively scheming to deceive and mislead the unwary.

 

The complex dynamic multivariate multidimensional violent nuance of twenty-two Formula One cars interacting with a race track surface for three hectic days will actively resist being reduced to any single simple percentage ... and that is as it should be. As it travels the world, any proper F1 car rightfully refuses to co-operate in being mischaracterized or synthesized by a one-line spreadsheet macro.

 

That's why, even with ready access to Excel, the teams and drivers all still insist on going to the inconvenient trouble of running all the races, just to make sure.

[Coops3]

^ So you think it's good then yea?

[RubalSher]

So I discarded the cumulative approach and rated each race on its own based on the ideas in this thread and here is how it looks from race to race.

 

[mark f1]

What data are you using for this RubalSher?  Its definitely not the team's fastest lap over the weekend, as the rest of the teams were not that close to Mercedes in China.

[RubalSher]

What data are you using for this RubalSher?  Its definitely not the team's fastest lap over the weekend, as the rest of the teams were not that close to Mercedes in China.

 

Sum of fastest laps of both drivers over the weekend. Mercedes sum is 197.041 seconds whereas the Ferrari for example is 197.739 seconds which is a gap of .35%.

[mark f1]

If you are comparing the pure performance of the team/car, then it would be better to use the fastest lap by either driver rather than the total of both.

[RubalSher]

If you are comparing the pure performance of the team/car, then it would be better to use the fastest lap by either driver rather than the total of both.

 

Can be done as you suggest but for now, but I will leave it as it is.

 

Eventually it is the drivers who drive these cars and driver factor has to be considered. It is possible any car could have gone faster or slower if there were different drivers. Taking both drivers in my opinion is more representative of "performance of the team-driver" combination as opposed to car alone.

[RubalSher]

If you are comparing the pure performance of the team/car, then it would be better to use the fastest lap by either driver rather than the total of both.

 

And if I did that, Ferrari in Bahrain would only be .05% behind Merc as opposed to the .35% they are shown above.

[RubalSher]

Also, qualy data will definitely be a step closer to the reality than practice data. Given that the only qualy data we have is from Bahrain, that may be a better indicator of the actual reality compared to the other 3 races.

[mark f1]

.14% using practice 2 fast laps for Lewis and Fernando.

[RubalSher]

.14% using practice 2 fast laps for Lewis and Fernando.

 

Yes .14%. I was just trying to guess without doing the actual maths.

[AlexS]

 

Sum of fastest laps of both drivers over the weekend. Mercedes sum is 197.041 seconds whereas the Ferrari for example is 197.739 seconds which is a gap of .35%.

 

That is not good enough, for example Kimi had a lap almost as fast as Alonso but due to tire issues he couldn't sustain that pace. It obviously also doesn't take care of cars that at end of the race are coasting and those that are fighting.

[AlexS]

The only way for a correct start to a study like this is to filter those laps that can correctly show the pace.

[Lazy]

That is not good enough, for example Kimi had a lap almost as fast as Alonso but due to tire issues he couldn't sustain that pace. It obviously also doesn't take care of cars that at end of the race are coasting and those that are fighting.

 

 

The only way for a correct start to a study like this is to filter those laps that can correctly show the pace.

Read through the thread, it's explained in detail.

[RubalSher]

 

Nice job, it was something like this which I had in mind [I would may "swap/invert" the y-axis, so that the "best" car/team is shown at the top, and everyone else below, but that's just a personal preference - up to you how to want to do it. It doesn't change the information content].

 

In parting, allow me to make two comments/suggestions [again, it's up to you, if you want to run with it or not]

- you could consider, showing the data as "discrete data points" instead of a line, and then fit a "trend line" [linear perhaps] to this data points.

  In doing so, you have the information how each race contributed to the "overall picture" , via the location of the data points, and the trendline would show the underlying tendency/trend as in how much teams gain or lose in relation to the "pace setting" team. The inclination of the line, will indicate how much the gain/lose, as an example if this line runs horizontal, it would indicate, that "on average" over the season (or the races shown so far) the team, keep the distance to the leader constant. it may came closer for one or two races (track/weather specific result perhaps) and on others, the gap has increased, but the trend line, would show the "general underlying tendency/trend" maybe a bit clearer / with less noise.

 

- if you want take both cars of each team into account, then a alternative approach would be to average them, instead of adding summing them, but again, that's mainly personal preference.

using both cars, brings the driver component more into the analysis (for better or worse it depends on what you want to get out of it), and the overall performance shown, doesn't only indicate how good a car a team has build, but would benefit teams which have two evenly matched drivers, if both are close to the performance limit of the car.

If this is "good or bad" again, depends on what you want to learn/see/get out of your analysis - different ways to skin a cat - some will agree with an particular approach, others may not, that's all good and fair IMHO. There will never be a "absolute" perfect way/model for any of this.

 

The value/accuracy of any such model, will increase over time (with more data points), it's still fairly early days, so "outliners" can have a larger effect in skewing the "general picture" one wants to paint.

Nevertheless, I think, it starts to show some "trends" which may are not so far from the truth, and also gives some indication of how much PU performance contributes to the overall car performance picture. I think, we should wait two races more, to see a "baseline" trend, and than it will interesting to see how this develops over the rest of the season. 

 

Have fun

 

Running short on time now but good ideas. Will be back in a couple of days maybe to try what you suggested.

[f1RacingForever]

We can all get a good sense of the pecking order from the races so all this analysis seems silly and unnecessary. I don't need a graph to tell me Ferrari are faster than Williams.

[mark f1]

Updated after Spain.

 

Before you ask, this is the rolling average over the whole season, taking each drivers fastest lap of the weekend as a percentage of the fastest lap for the weekend. It aims to show how each driver/car combination compares in ultimate pace over the entire season.

 

Pos  Driver        Team          Average

1    Rosberg       Mercedes      100.123%

2    Hamilton      Mercedes      100.163%

4    Ricciardo     Red Bull      101.050%

3    Alonso        Ferrari       101.193%

5    Vettel        Red Bull      101.375%

6    Raikkonen     Ferrari       101.420%

7    Massa         Williams      101.480%

9    Bottas        Williams      101.540%

8    Button        McLaren       101.572%

10   Hulkenberg    Force India   101.892%

11   Magnussen     McLaren       101.976%

12   Perez         Force India   102.102%

13   Vergne        Toro Rosso    102.174%

14   Kvyat         Toro Rosso    102.360%

15   Guiterrez     Sauber        102.889%

17   Grosjean      Lotus         102.906%

16   Sutil         Sauber        103.045%

18   Maldonado     Lotus         103.919%

19   Bianchi       Marussia      104.656%

20   Chilton       Marussia      105.000%

21   Kobayashi     Caterham      105.574%

22   Ericcson      Caterham      105.672%

 

Comments:

• Rosberg still maintains the lead but only just now.
• A number of drivers had their fastest in practice 2.
• Only 3 drivers improved their gap to the ultimate pace. Hamilton, of course as he had the fastest lap, and both Lotus drivers. So Lotus were the only team to reduce their gap to the Mercedes.
• For the first time, every driver excluding both Mercedes were more than 1% slower than the ultimate Mercedes pace.
• Overtakes – Ricciardo overtakes Alonso, Bottas over Button, Grosjean over Sutil
• Whilst Bottas had his best qualifying position in fourth, he was further away from Mercedes on pace than he was in Bahrain and China

        

 

 

[RubalSher]

[Juan Kerr]

So I discarded the cumulative approach and rated each race on its own based on the ideas in this thread and here is how it looks from race to race.

 

That's a great illustration but means nothing to people like me who are colourblind.

[mark f1]

 

I have the same graph at a driver level, but there are too many lines close together to make it readable.

[Sennasational]

So if it's dry for FP1, and lets say Caterham get out early to do some low fuel runs on option tyres, climbing to the top of the time sheets. And then it starts raining and the conditions remain wet for the rest of the weekend. Does this mean that the Caterhams would be the fastest car that weekend? According to this method of analysis it would.

 

Qualifying neutralizes most variables for every team apart from the occasional bit of traffic. You have every driver pushing as hard as they can, with the least amount of fuel possible, on a brand new set of the fastest tyres available, all driving on the same track, at the same time, in the same weather. That is the point of qualifying, we see who is fastest over a single lap. Then in the race, we find out who is fastest over 2 hours of laps. What information does this give us that a quick look at the results tables don't show?

 

I fail to see how this provides any worthwhile information whatsoever. Maybe that's just me.

 

EDIT: By the way, this isn't supposed to be a critical rant at what you're doing, i applaud your efforts to analyse. I'm simply critisizing Gary Anderson's logic. However, if this is the data he uses to make his 'predictions' then it sure does explain why he's always so hopelessly and hilariously wong. About everything.

Edited by Sennasational, 15 May 2014 - 00:51.

[lbennie]

That's a great illustration but means nothing to people like me who are colourblind.

 

Atleast hes putting the effort in. no need to be a douche about it 

[Paco]

Seriously people.. how can people try and discuss flawed meaningless data. Especially, in a season of dominance where it's highly doubtful either merc driver is ever pushing to the limit over practice.. so really only qualifying comparison and only if done on the same q3 and only if same tire type... otherwards.. near impossible. stupid idea and thread. Maybe if this was mid 90s with all out qualifying...

[RubalSher]

[RubalSher]